The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several bundled scripts to perform its operations, including scripts/anysearch_cli.py, scripts/anysearch_cli.js, scripts/anysearch_cli.ps1, and scripts/anysearch_cli.sh. The agent is instructed to select the appropriate runtime (Python, Node.js, etc.) and execute these scripts with search queries and parameters.
[DATA_EXFILTRATION]: Search queries, domain-specific parameters, and target URLs for extraction are sent to the vendor's API endpoint at https://api.anysearch.com/mcp. This communication is the primary purpose of the skill and is documented as using JSON-RPC over HTTPS. API keys are handled via the Authorization: Bearer header.
[PROMPT_INJECTION]: The SKILL.md file contains detailed logic for the agent to follow, such as prioritizing vertical searches over general ones and handling 'auto-registered' API keys. These instructions are designed to improve tool performance and manage service quotas rather than bypass security controls.
[EXTERNAL_DOWNLOADS]: The README.md provides installation instructions that involve downloading the skill package from the vendor's GitHub repository (anysearch-ai/anysearch-skill). This is standard distribution behavior for third-party skills. The Python implementation also lists requests as a dependency in requirements.txt.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes data from external web sources.
Ingestion points: The extract and search commands in all CLI scripts fetch content and snippets from arbitrary external URLs.
Boundary markers: None explicitly implemented within the script output; the agent relies on its own system prompt and the context provided in SKILL.md.
Capability inventory: The skill has the capability to write to the .env file (for API keys) and runtime.conf (for state) based on instructions in SKILL.md, and it can execute bundled shell scripts.
Sanitization: The CLI scripts act as a pass-through for the API's JSON results and do not perform content sanitization on the retrieved text.

anysearch